On Mon, 2 Feb 2004 20:42:50 +1100
[EMAIL PROTECTED] wrote:

> On Mon, 2 Feb 2004 19:58:15 +1100
> Erik de Castro Lopo <[EMAIL PROTECTED]> wrote:
> >  [ ... ]
> > As good as OO is for files, the idea the everything should be an object
> > in a class hierarchy is just as wrong.
> 
> I agree with you and from what I read of Stroustrup, he
> would agree with you too.
> 
> > Personally I think Stroustrup has done more damage to the computer
> > industry that Bill Gates.
> 
> Which makes that statement ludicrous.  Holy hyperbole batman.

Maybe comparing him to Bill Gates was a little unfair. I therefore
nominate him as the person who has done the third most damage to
the computer industry, after Bill Gates and Microsoft.

However, the benchmarketing below would make any microsoft 
marketroid proud.

> Have a look at these simple examples in C and C++,
> and tell me that C++ is not worth it:
> 
>   http://www.research.att.com/~bs/new_learning.pdf

That is one of the funniest things I have read in a long time. The
opening sentence "We want to make our programs easy to write, correct,
maintainable and acceptably efficient." from the main architect of 
C++. The irony is astounding.

The toy example programs might be fine for C++ but the C programs are 
good candidates for the Strawman of the Year award. In the first 
program he uses scanf() which most experienced C coders ignore; 
fgets() is a far better choice. He then tries to replace scanf with
some complete garbage.

He then goes on to an equally bogus efficiency example. Last time 
someone showed me stupid speed comparisons like this, I disassembled 
the C and C++ binaries to prove that g++ was inlining non-inline 
functions while gcc was not. By adding the C99 inline directive to 
the C program the two programs compiled to almost identical binaries 
with zero measurable performance difference.

As for the comment on Stroustrup, I will happily repeat it. I first
learnt C++ in about 1990 but grew disillusion by it by about 1994.
I do however, still use for the rare task where C++ actually fits
the problem better than C or Python.

The constant changes to the C++ language standard, caused millions 
of man hours spent chasing the C++ standards and the constantly 
changing ABIs caused similar amounts of time recompiling and 
debugging code and sold almost as many disk drives as windows.
The C++ zealotry that was so prevalent through much of the nineties 
caused endless re-writes of perfectly good, working C code. A 
generation of coders had the bogus class heirarchy examples (that 
everybody now realises were bogus) shoved down their throats. Not 
to mention the countless hundreds of millions of hours have been 
spent trying to decipher the meanings of some of the more convoluted 
gunk that C++ allows and even encourages.

A really good example of how messed up C++ is, is that backbone of
the C++ cannon; iostream (attributable to Stroustrup I believe). 
Iostream is so ugly that even its parent couldn't love it. It does 
however serve as a glowing example of the abuse of operater 
overloading.

All in all, I believe that C++ was released to the public a good
decade before it was ready for public consumption.

Erik
-- 
+-----------------------------------------------------------+
  Erik de Castro Lopo  [EMAIL PROTECTED] (Yes it's valid)
+-----------------------------------------------------------+
"If you are going to SCO to license superior driver technology,
 you need to reevaluate your product."
  -- Comment about SUN and SCO on Linuxtoday.com
-- 
SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/
Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html

Reply via email to